Since the introduction of the first neural Text-to-SQL systems, a multitude of different approaches have been proposed, each trying to advance the state of the art in its own way. Understanding the trends and directions of this research area by looking at all the different systems can often be like looking for a needle in a haystack. However, when we systematically analyze all the different techniques that have been used, we start to observe the commonalities of different systems and the emerging trends. This analysis and categorization help us better understand the work that has been done and the possible design choices for a Text-to-SQL system. This chapter presents an architecture for neural Text-to-SQL systems that covers all the design choices that have been presented in the literature. We present all the parts of a system starting from the given inputs, until the generation of a SQL query and even possible refinements after a prediction has been made. For each step in the process we analyze the different possible choices, explaining their differences, advantages, and drawbacks, in an effort to provide a clearer picture of the research landscape.